Limits...
Forest Walk Methods for Localizing Body Joints from Single Depth Image.

Jung HY, Lee S, Heo YS, Yun ID - PLoS ONE (2015)

Bottom Line: A regression forest is trained to estimate the probability distribution to the direction or offset toward the particular joint, relative to the adjacent position.The distribution for next position is found from traversing the regression tree from new position.The continual position sampling through 3D space will eventually produce an expectation of sample positions, which we estimate as the joint position.

View Article: PubMed Central - PubMed

Affiliation: Division of Computer and Electronic Systems Engineering, Hankuk University of Foreign Studies, Yongin, Republic of Korea.

ABSTRACT
We present multiple random forest methods for human pose estimation from single depth images that can operate in very high frame rate. We introduce four algorithms: random forest walk, greedy forest walk, random forest jumps, and greedy forest jumps. The proposed approaches can accurately infer the 3D positions of body joints without additional information such as temporal prior. A regression forest is trained to estimate the probability distribution to the direction or offset toward the particular joint, relative to the adjacent position. During pose estimation, the new position is chosen from a set of representative directions or offsets. The distribution for next position is found from traversing the regression tree from new position. The continual position sampling through 3D space will eventually produce an expectation of sample positions, which we estimate as the joint position. The experiments show that the accuracy is higher than current state-of-the-art pose estimation methods with additional advantage in computation time.

No MeSH data available.


Related in: MedlinePlus

(a) mAP using 1 tree. (b) mAP using 2 tree forest. (c) mAP using 3 tree forest. (d) Average fps. In (a) (b) and (c), mAPs of proposed pose estimation algorithms are shown in respect to the number of steps and jumps taken. An additional accuracy gain can be observed for an increased tree number in random forest. (d) shows the average fps for different number of samples and trees.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4581738&req=5

pone.0138328.g005: (a) mAP using 1 tree. (b) mAP using 2 tree forest. (c) mAP using 3 tree forest. (d) Average fps. In (a) (b) and (c), mAPs of proposed pose estimation algorithms are shown in respect to the number of steps and jumps taken. An additional accuracy gain can be observed for an increased tree number in random forest. (d) shows the average fps for different number of samples and trees.

Mentions: The proposed RFW, GFW, RFJ, and GFJ methods are compared across different tree numbers in the forest. See Fig 5. For all four methods, the inclusion of additional trees in the forest increased the mAP as expected. The highest precisions are consistently obtained by GFW algorithm. RFW algorithm achieved the similar mAP to that of GFW when using only 1 tree. However, when the number of trees is increased in the forest, GFW algorithm has notably higher mAP than RFW as shown in Fig 5c. As the number of trees increases, the accuracy of random forest output also increases, and thus it becomes more sensible to trust the average direction than to randomly select direction from the distribution. Otherwise, mAP of RFW and GFW converge to the same mAP as the number of steps become higher.


Forest Walk Methods for Localizing Body Joints from Single Depth Image.

Jung HY, Lee S, Heo YS, Yun ID - PLoS ONE (2015)

(a) mAP using 1 tree. (b) mAP using 2 tree forest. (c) mAP using 3 tree forest. (d) Average fps. In (a) (b) and (c), mAPs of proposed pose estimation algorithms are shown in respect to the number of steps and jumps taken. An additional accuracy gain can be observed for an increased tree number in random forest. (d) shows the average fps for different number of samples and trees.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4581738&req=5

pone.0138328.g005: (a) mAP using 1 tree. (b) mAP using 2 tree forest. (c) mAP using 3 tree forest. (d) Average fps. In (a) (b) and (c), mAPs of proposed pose estimation algorithms are shown in respect to the number of steps and jumps taken. An additional accuracy gain can be observed for an increased tree number in random forest. (d) shows the average fps for different number of samples and trees.
Mentions: The proposed RFW, GFW, RFJ, and GFJ methods are compared across different tree numbers in the forest. See Fig 5. For all four methods, the inclusion of additional trees in the forest increased the mAP as expected. The highest precisions are consistently obtained by GFW algorithm. RFW algorithm achieved the similar mAP to that of GFW when using only 1 tree. However, when the number of trees is increased in the forest, GFW algorithm has notably higher mAP than RFW as shown in Fig 5c. As the number of trees increases, the accuracy of random forest output also increases, and thus it becomes more sensible to trust the average direction than to randomly select direction from the distribution. Otherwise, mAP of RFW and GFW converge to the same mAP as the number of steps become higher.

Bottom Line: A regression forest is trained to estimate the probability distribution to the direction or offset toward the particular joint, relative to the adjacent position.The distribution for next position is found from traversing the regression tree from new position.The continual position sampling through 3D space will eventually produce an expectation of sample positions, which we estimate as the joint position.

View Article: PubMed Central - PubMed

Affiliation: Division of Computer and Electronic Systems Engineering, Hankuk University of Foreign Studies, Yongin, Republic of Korea.

ABSTRACT
We present multiple random forest methods for human pose estimation from single depth images that can operate in very high frame rate. We introduce four algorithms: random forest walk, greedy forest walk, random forest jumps, and greedy forest jumps. The proposed approaches can accurately infer the 3D positions of body joints without additional information such as temporal prior. A regression forest is trained to estimate the probability distribution to the direction or offset toward the particular joint, relative to the adjacent position. During pose estimation, the new position is chosen from a set of representative directions or offsets. The distribution for next position is found from traversing the regression tree from new position. The continual position sampling through 3D space will eventually produce an expectation of sample positions, which we estimate as the joint position. The experiments show that the accuracy is higher than current state-of-the-art pose estimation methods with additional advantage in computation time.

No MeSH data available.


Related in: MedlinePlus